World Disaster Risk Analysis¶

In [1]:
#!conda install geopandas -y
In [37]:
import numpy as np
import pandas as pd

%matplotlib inline
import matplotlib.pyplot as plt
#plots.style.use('fivethirtyeight')

import math as mt
from scipy.stats import norm
import seaborn as sns

import plotly.express as px
import plotly.graph_objects as go
import geopandas as gpd

Import Data¶

In [3]:
wri = pd.read_csv('Data/world_risk_index.csv')
wri.head()
Out[3]:
Region WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category
0 Vanuatu 32.00 56.33 56.81 37.14 79.34 53.96 2011 Very High Very High High High
1 Tonga 29.08 56.04 51.90 28.94 81.80 44.97 2011 Very High Very High Medium Medium
2 Philippinen 24.32 45.09 53.93 34.99 82.78 44.01 2011 Very High Very High High High
3 Salomonen 23.51 36.40 64.60 44.11 85.95 63.74 2011 Very High Very High Very High High
4 Guatemala 20.88 38.42 54.35 35.36 77.83 49.87 2011 Very High Very High High High

Background:¶

The WorldRiskIndex is a statistical model that provides an assessment of the latent risk of 193 countries falling victim to a humanitarian disaster caused by extreme natural events and the negative impacts of climate change. Based on peer-reviewed concepts of risk, hazard and vulnerability, it is assumed that disaster risks are not solely shaped by the occurrence, intensity, and duration of extreme natural events, but that social factors, political conditions, and economic structures are equally responsible for whether disasters occur in the context of extreme natural events. Accordingly, both main spheres of disaster risk, exposure and vulnerability, are treated as equals.

The WorldRiskIndex was initially developed in 2011 by the United Nations University Institute for Environment and Human Security (UNU-EHS) for Bündnis Entwicklung Hilft as a model with 27 indicators to analytically link and relate the two spheres of disaster risks – exposure to natural hazards such as earthquakes, storms or droughts, and societal capacities to respond to these kinds of events. The methodology of the WorldRiskIndex has been continuously revised and developed by the Institute for International Law of Peace and Armed Conflict (IFHV) since 2018. In 2022, a new, fully revised model of the WorldRiskIndex was published, enabling more accurate analyses by incorporating more than 100 high-quality indicators, new data sources, and more robust statistical methods, thus finally replacing the previously used model.

source: https://data.humdata.org/dataset/worldriskindex

Description of the data:¶

  • Region: Name of the region.
  • WRI: World Risk Score of the region.
  • Exposure: Risk/exposure to natural hazards such as earthquakes, hurricanes, floods, droughts, and sea level rise.
  • Vulnerability: Vulnerability depending on infrastructure, nutrition, housing situation, and economic framework conditions.
  • Susceptibility: Susceptibility depending on infrastructure, nutrition, housing situation, and economic framework conditions.
  • Lack of Coping Capabilities: Coping capacities in dependence of governance, preparedness and early warning, medical care, and social and material security.
  • Lack of Adaptive Capacities: Adaptive capacities related to coming natural events, climate change, and other challenges.
  • Year: Year data is being described.
  • WRI Category: WRI Category for the given WRI Score.
  • Exposure Category: Exposure Category for the given Exposure Score.
  • Vulnerability Category: Vulnerability Category for the given Vulnerability Score.
  • Susceptibility Category: Susceptibility Category for the given Susceptibility Score.

source: https://www.kaggle.com/datasets/tr1gg3rtrash/global-disaster-risk-index-time-series-dataset/data

Observations:¶

The data collected spans from 2011 to 2021. Based on the description, the two primary spheres of disaster risk are exposure and vulnerability, which are treated as equally important.

  • Exposure is influenced by the occurrence, intensity, and duration of extreme natural events.
  • Vulnerability is shaped by social factors, political conditions, and economic structures.

It will be interesting for us to explore the WRI of countries, with a focus on its relationship exposure and vulnerability.

Data Cleaning¶

In [4]:
wri.isnull().sum()
Out[4]:
Region                          0
WRI                             0
Exposure                        0
Vulnerability                   0
Susceptibility                  0
Lack of Coping Capabilities     0
 Lack of Adaptive Capacities    1
Year                            0
Exposure Category               0
WRI Category                    1
Vulnerability Category          4
Susceptibility Category         0
dtype: int64
In [5]:
wri[wri.duplicated()]
Out[5]:
Region WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category
In [6]:
wri.rename(columns={' Lack of Adaptive Capacities': 'Lack of Adaptive Capacities'}, inplace=True)
wri.rename(columns={'Region': 'Country'}, inplace=True)
wri.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1917 entries, 0 to 1916
Data columns (total 12 columns):
 #   Column                       Non-Null Count  Dtype  
---  ------                       --------------  -----  
 0   Country                      1917 non-null   object 
 1   WRI                          1917 non-null   float64
 2   Exposure                     1917 non-null   float64
 3   Vulnerability                1917 non-null   float64
 4   Susceptibility               1917 non-null   float64
 5   Lack of Coping Capabilities  1917 non-null   float64
 6   Lack of Adaptive Capacities  1916 non-null   float64
 7   Year                         1917 non-null   int64  
 8   Exposure Category            1917 non-null   object 
 9   WRI Category                 1916 non-null   object 
 10  Vulnerability Category       1913 non-null   object 
 11  Susceptibility Category      1917 non-null   object 
dtypes: float64(6), int64(1), object(5)
memory usage: 179.8+ KB
In [7]:
wri.isnull().sum()
Out[7]:
Country                        0
WRI                            0
Exposure                       0
Vulnerability                  0
Susceptibility                 0
Lack of Coping Capabilities    0
Lack of Adaptive Capacities    1
Year                           0
Exposure Category              0
WRI Category                   1
Vulnerability Category         4
Susceptibility Category        0
dtype: int64

To handle the missing "Category" values, we will fill in the missing values based on the data. For the quantitative data, we will replace missing values with the mean.

In [8]:
#show the location of the nulls
rows_with_nulls = wri[wri.isnull().any(axis=1)]
rows_with_nulls
Out[8]:
Country WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category
1193 Österreich 2.87 13.18 21.75 13.63 39.27 12.34 2019 Medium Very Low NaN Very Low
1202 Deutschland 2.43 11.51 21.11 14.30 36.44 12.60 2019 Low Very Low NaN Very Low
1205 Norwegen 2.34 10.60 22.06 13.29 39.21 13.68 2019 Low Very Low NaN Very Low
1292 Föd. Staaten v. Mikronesien 7.59 14.95 50.77 31.79 72.13 48.39 2020 High NaN High High
1858 Korea Republic of 4.59 14.89 30.82 14.31 46.55 31.59 NaN 2016 Very High Very High NaN High
In [9]:
#Calculate Min and Max Values for Each Category
data_2019 = wri[wri['Year'] == 2019]
data_2020 = wri[wri['Year'] == 2020]

category_stats = data_2019.groupby('Vulnerability Category')['Vulnerability'].agg(['min', 'max']).reset_index() #to dataframe
category_stats = category_stats.sort_values(by='min')
category_stats
Out[9]:
Vulnerability Category min max
4 Very Low 22.49 33.01
1 Low 33.10 41.98
2 Medium 42.18 47.86
0 High 47.98 61.74
3 Very High 61.99 76.13
In [10]:
#Fill in the missing values
wri.loc[1193,'Vulnerability Category'] = 'Very Low'
wri.loc[1202,'Vulnerability Category'] = 'Very Low'
wri.loc[1205,'Vulnerability Category'] = 'Very Low'
In [11]:
#Calculate Min and Max Values for Each Category
data_2016 = wri[wri['Year'] == 2016]
category_stats = data_2016.groupby('Vulnerability Category')['Vulnerability'].agg(['min', 'max']).reset_index() 
category_stats = category_stats.sort_values(by='min')
category_stats
Out[11]:
Vulnerability Category min max
4 Very Low 24.79 34.40
1 Low 35.10 43.11
2 Medium 43.40 49.72
0 High 50.57 62.58
3 Very High 63.01 74.80
In [12]:
wri.loc[1858,'Vulnerability Category'] = 'Very Low'
In [13]:
#Calculate Min and Max Values for Each Category
data_2020 = wri[wri['Year'] == 2020]
category_stats = data_2020.groupby('WRI Category')['WRI'].agg(['min', 'max']).reset_index() 
category_stats = category_stats.sort_values(by='min')
category_stats
Out[13]:
WRI Category min max
4 Very Low 0.31 3.14
1 Low 3.30 5.66
2 Medium 5.68 7.57
0 High 7.71 10.51
3 Very High 10.76 49.74
In [14]:
#Fill in the missing value
wri.loc[1292,'WRI Category'] = 'Medium'
In [15]:
mean_lack_adaptive = data_2016['Lack of Adaptive Capacities'].mean()
wri.loc[1858,'Lack of Adaptive Capacities']=mean_lack_adaptive
In [16]:
wri.isnull().sum()
Out[16]:
Country                        0
WRI                            0
Exposure                       0
Vulnerability                  0
Susceptibility                 0
Lack of Coping Capabilities    0
Lack of Adaptive Capacities    0
Year                           0
Exposure Category              0
WRI Category                   0
Vulnerability Category         0
Susceptibility Category        0
dtype: int64

We have cleaned our dataset, making sure there are no null or duplicate values.

Note: Some of the missing category values fall outside the observed range, suggesting they may have been treated as outliers. Furthermore, filling in missing values for "Lack of Adaptive Capacities" with the mean could introduce bias. If we do not use models that rely on a complete dataset, it might be more appropriate to drop these rows with missing values to avoid potential distortion of analysis.

Exploratory Data Analysis¶

In [17]:
wri.describe()
Out[17]:
WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year
count 1917.000000 1917.000000 1917.000000 1917.000000 1917.000000 1917.000000 1917.000000
mean 7.551763 15.388336 48.075759 30.739431 70.438289 43.090562 2016.049557
std 5.553257 10.240135 13.835666 15.667030 15.038854 13.547619 3.182045
min 0.020000 0.050000 14.310000 8.260000 31.590000 11.160000 2011.000000
25% 3.740000 10.160000 37.040000 17.790000 59.330000 33.200000 2013.000000
50% 6.520000 12.760000 47.100000 25.400000 74.230000 43.090000 2016.000000
75% 9.400000 16.450000 60.060000 42.640000 83.000000 53.060000 2019.000000
max 56.710000 99.880000 76.470000 70.830000 94.360000 76.110000 2021.000000
In [18]:
quantitative_features = ['WRI', 'Exposure', 'Vulnerability', 'Susceptibility', 
                         'Lack of Coping Capabilities', 'Lack of Adaptive Capacities']

# Calculate the correlation matrix
correlation_matrix = wri[quantitative_features].corr()
In [19]:
# Create the heatmap
plt.figure(figsize=(5, 4))
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', vmin=-1, vmax=1, square=True)
plt.title('Correlation Heatmap')
plt.show()

The correlation heatmap provides a visual representation of the strength and direction of the relationships between the six quantitative features: WRI, Exposure, Vulnerability, Susceptibility, Lack of Coping Capabilities, and Lack of Adaptive Capacities.

Key Observations¶

WRI and Exposure 0.92: There is a very strong positive correlation between WRI and Exposure. This suggests that exposure is a heavily weighted in determining determing the world risk index.

WRI and Vulnerability 0.44: There is a moderate positive correlation between WRI and Vulnerability. While Vulnerability is associated with WRI, its influence is less significant compared to Exposure.

Vulnerability and Susceptibility, Lack of Adaptive Capacities, and Lack of Coping Capabilities: 0.93, 0.94, 0.93: There is a similarly strong positive correlation between Vulnerability and the factors of Susceptibility, Lack of Adaptive Capacities, and Lack of Coping Capabilities. "Societal vulnerability is divided into susceptibility to extreme natural events, lack of coping capacities, and lack of adaptive capacities." This suggests that Vulnerability is significantly influenced by these three sub-components.

Susceptibility, Lack of Adaptive Capacities, and Lack of Coping Capabilities 0.78, 0.83, 0.84: The heat graph shows that the each of the three measures are strongly correlated to the others. This suggests that regions struggling with one are likely to struggle with the other, compounding their overall vulnerability.

Exposure and Vulnerability 0.10: The low positive correlation between Exposure and Vulnerability suggests that these two variables are relatively independent. Regions with high exposure do not necessarily have high vulnerability, and vice versa.

Summary:¶

We have observed that the World Risk Index (WRI) is primarily related to Exposure, rather than Vulnerability. Additionally, Vulnerability is significantly influenced by its sub-components, such as Susceptibility, Lack of Coping Capacities, and Lack of Adaptive Capacities. The strong correlations among these sub-components suggest that some regions face compounded challenges, which could heighten their vulnerability and make recovery and improvements more difficult.

In [20]:
mean_wri = wri.groupby('Country').mean().reset_index()

fig = px.scatter(mean_wri, 
                 x='Exposure',  
                 y='Vulnerability',       
                 color='WRI', 
                #text='Region',  
                 title='Scatter Plot of Vulnerability vs Exposure',
                 labels={'Vulnerability': 'Vulnerability', 'Exposure': 'Exposure', 'Country':'Country'})

fig.update_layout(
    width=1000, 
    height=800, 
    xaxis_title='Exposure',
    yaxis_title='Vulnerability'      
)

fig.show()
In [21]:
year_group = wri.groupby('Year').mean()
In [ ]:
year_group_reset = year_group.reset_index()

fig = px.line(year_group_reset, x='Year', y=year_group.columns, labels={'value': 'Value'}, title='Line Graphs of Various Indicators Over Time')

fig.show()

Overall, Vulnerability and its subcategories of Lack of Adaptive Capacities and Susceptibility have been on a downward trend from 2011 to 2021, indicating potential improvements in these areas over time. Meanwhile, Lack of Coping Capabilities, Exposure, and WRI show slight increases.

In [ ]:
data_2021 = wri[wri['Year'] == 2021]
wri_2021 = data_2021[['Country','WRI']]
gdf = gpd.read_file(gpd.datasets.get_path('naturalearth_lowres'))
gdf = gdf.rename(columns={'name': 'Country'})
gapminder = px.data.gapminder()
In [ ]:
#gdf_2021 = pd.merge(gdf, wri_2021, how='left', left_on='Country', right_on='Country')
df = pd.merge(gapminder, wri_2021, how='left', left_on='country', right_on='Country')

fig = px.choropleth(df, locations="iso_alpha",
                    color="WRI", 
                    hover_name="Country", 
                    color_continuous_scale=px.colors.sequential.Plasma,
                    title='World Risk Index (WRI) by Country for 2021',
                    width=1000,  
                    height=600)

np.random.seed(12)
fig.show()
In [ ]:
us_data = wri[wri['Country'] == 'United States']
us_data
Out[ ]:
Country WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category
643 United States 3.88 12.25 31.67 16.47 48.57 29.98 2014 Medium Low Very Low Very Low
1872 United States 3.76 12.25 30.68 16.35 48.24 27.46 2016 Medium Low Very Low Very Low

This map shows that some countries in Africa, South America and Central America have the highest disaster risk in 2021. The table below confirms that the blank areas on the map correspond to missing data. In the next map, we will use the mean WRI values over 2011 to 2021 for each country to provide a more comprehensive view of disaster risk across countries.

In [ ]:
mean_wri = wri.groupby('Country').mean().reset_index()
df = pd.merge(gapminder, mean_wri, how='left', left_on='country', right_on='Country')

fig = px.choropleth(df, locations="iso_alpha",
                    color="WRI", 
                    hover_name="Country", 
                    color_continuous_scale=px.colors.sequential.Plasma,
                    title='Average World Risk Index (WRI) by Country',
                    width=1000,  
                    height=600)

np.random.seed(12)
fig.show()

This choropleth map highlights that certain countries in West Africa, Southeast Asia, and Central America have the highest World Risk Index (WRI). Notably, the Philippines stands out with the highest WRI on the map at 27.47. In the earlier map, the Philippines was not included in the data, causing other countries to appear with disproportionately high WRIs compared to their 10-year averages. However, by hovering over the countries on this map, we can observe that most have not experienced significant increases in WRI.

Now, let’s take a closer look at some of the countries with the highest WRI individually.

In [ ]:
regions_of_interest = ['Guatemala', 'Honduras', 'Nicaragua', 'Niger', 'Nigeria', 'Chile', 'Uruguay']

wri_filtered = wri[wri['Country'].isin(regions_of_interest)]

wri_pivoted_wri = wri_filtered.pivot(index='Year', columns='Country', values='WRI')
wri_pivoted_exposure = wri_filtered.pivot(index='Year', columns='Country', values='Exposure')

fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(20, 8))

# WRI
for region in regions_of_interest:
    axes[0].plot(wri_pivoted_wri.index, wri_pivoted_wri[region], label=region)

axes[0].set_title('WRI Over Time for Selected Regions')
axes[0].set_xlabel('Year')
axes[0].set_ylabel('WRI')
axes[0].legend(title='Country')
axes[0].grid(True)

# Exposure
for region in regions_of_interest:
    axes[1].plot(wri_pivoted_exposure.index, wri_pivoted_exposure[region], label=region)

axes[1].set_title('Exposure Over Time for Selected Regions')
axes[1].set_xlabel('Year')
axes[1].set_ylabel('Exposure')
axes[1].legend(title='Country')
axes[1].grid(True)

plt.tight_layout()
plt.show()

Since 2018, Niger, Nigeria, and Uruguay have seen both an increase in their World Risk Index (WRI) and Exposure.

In [ ]:
regions_of_interest = ['Guatemala', 'Honduras', 'Nicaragua', 'Niger', 'Nigeria', 'Chile', 'Uruguay']

num_regions = len(regions_of_interest)
rows = (num_regions + 1) // 2  
fig, axes = plt.subplots(rows, 2, figsize=(14, 5 * rows))

axes = axes.flatten()

for i, region in enumerate(regions_of_interest):
    wri_region_yr = wri[wri['Country'] == region].groupby('Year').mean()
    
    for column in wri_region_yr.columns:
        axes[i].plot(wri_region_yr.index, wri_region_yr[column], marker='o', label=column)
    
    axes[i].set_xlabel('Year')
    axes[i].set_ylabel('Value')
    axes[i].set_title(f'Line Graphs of Various Indicators Over Time for {region}')
    axes[i].legend(loc='best')
    axes[i].grid(True)

if num_regions % 2 != 0:
    fig.delaxes(axes[-1])

plt.tight_layout()
plt.show()

Importing Country and HDI Data¶

In [ ]:
countries = pd.read_csv('Data/Countries.csv')
#countries.head()
df = pd.merge(countries, wri, how='left', on='Country')
df.head()
Out[ ]:
Country Code Continent Region WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category
0 Afghanistan AFG Asia Southern Asia 14.06 18.45 76.19 61.09 93.94 73.55 2011.0 Very High Very High Very High Very High
1 Afghanistan AFG Asia Southern Asia 9.93 13.17 75.41 56.67 93.44 76.11 2013.0 Medium High Very High Very High
2 Afghanistan AFG Asia Southern Asia 9.55 13.17 72.49 55.77 92.36 69.33 2015.0 Medium High Very High Very High
3 Afghanistan AFG Asia Southern Asia 9.71 13.17 73.73 55.93 93.37 71.89 2014.0 Medium High Very High Very High
4 Afghanistan AFG Asia Southern Asia 8.18 12.27 66.63 48.57 91.40 59.93 2021.0 Medium High Very High Very High
In [ ]:
df['Region'].nunique()
Out[ ]:
23
In [ ]:
#gdf = gpd.read_file(gpd.datasets.get_path('naturalearth_lowres'))
#gdf = gdf.rename(columns={'name': 'Country'})
#gdf = gdf.merge(wri, on='Country', how='left')
#gdf.head()
In [ ]:
HDI_2021 = pd.read_csv('Data/2021_development.csv')
In [ ]:
WRI_2021 = wri[wri['Year'] == 2021]
wri_hdi_2021 = pd.merge(WRI_2021, HDI_2021, how='left', on='Country')
wri_hdi_2021.head()
Out[ ]:
Country WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category HDI rank Human Development Index (HDI) - 2021 Life expectancy at birth - 2021 Expected years of schooling - 2021 Mean years of schooling - 2021 Gross national income (GNI) per capita - 2021
0 Vanuatu 47.73 82.55 57.82 39.66 81.21 52.59 2021 Very High Very High High High 140.0 0.607 70.4 11.5 7.1 3085.0
1 Salomonen 31.16 51.13 60.95 46.07 81.14 55.63 2021 Very High Very High High Very High NaN NaN NaN NaN NaN NaN
2 Tonga 30.51 63.63 47.95 28.42 79.81 35.62 2021 Very High Very High Medium High 91.0 0.745 71.0 16.0 11.4 6822.0
3 Dominica 27.42 61.74 44.41 23.42 71.13 38.67 2021 Very High Very High Medium Medium 102.0 0.720 72.8 13.3 8.1 11488.0
4 Antigua und Barbuda 27.28 67.73 40.28 23.80 64.41 32.62 2021 Very High Very High Low Medium NaN NaN NaN NaN NaN NaN
In [ ]:
null_counts = wri_hdi_2021.isnull().sum()
null_counts
Out[ ]:
Country                                            0
WRI                                                0
Exposure                                           0
Vulnerability                                      0
Susceptibility                                     0
Lack of Coping Capabilities                        0
Lack of Adaptive Capacities                        0
Year                                               0
Exposure Category                                  0
WRI Category                                       0
Vulnerability Category                             0
Susceptibility Category                            0
HDI rank                                         111
Human Development Index (HDI) - 2021             111
Life expectancy at birth - 2021                  111
Expected years of schooling - 2021               111
Mean years of schooling - 2021                   111
Gross national income (GNI) per capita - 2021    111
dtype: int64
In [ ]:
wri_hdi_2021_cleaned = wri_hdi_2021.dropna(subset=['HDI rank'])
wri_hdi_2021_cleaned
Out[ ]:
Country WRI Exposure Vulnerability Susceptibility Lack of Coping Capabilities Lack of Adaptive Capacities Year Exposure Category WRI Category Vulnerability Category Susceptibility Category HDI rank Human Development Index (HDI) - 2021 Life expectancy at birth - 2021 Expected years of schooling - 2021 Mean years of schooling - 2021 Gross national income (GNI) per capita - 2021
0 Vanuatu 47.73 82.55 57.82 39.66 81.21 52.59 2021 Very High Very High High High 140.0 0.607 70.4 11.5 7.1 3085.0
2 Tonga 30.51 63.63 47.95 28.42 79.81 35.62 2021 Very High Very High Medium High 91.0 0.745 71.0 16.0 11.4 6822.0
3 Dominica 27.42 61.74 44.41 23.42 71.13 38.67 2021 Very High Very High Medium Medium 102.0 0.720 72.8 13.3 8.1 11488.0
5 Brunei Darussalam 22.77 58.17 39.14 15.33 68.13 33.96 2021 Very High Very High Low Very Low 51.0 0.829 74.6 14.0 9.2 64490.0
6 Guyana 21.83 43.93 49.69 25.96 77.23 45.88 2021 Very High Very High High Medium 108.0 0.714 65.7 12.5 8.6 22465.0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
161 Belarus 2.64 8.00 32.96 16.68 56.36 25.84 2021 Very Low Very Low Very Low Very Low 60.0 0.808 72.4 15.2 12.1 18849.0
163 Oman 2.54 6.04 42.02 23.68 66.65 35.73 2021 Very Low Very Low Low Medium 54.0 0.816 72.5 14.6 11.7 27054.0
175 Barbados 1.37 3.61 37.96 20.66 60.11 33.12 2021 Very Low Very Low Low Low 70.0 0.790 77.6 15.7 9.9 12306.0
176 Grenada 1.06 2.40 43.98 26.36 69.21 36.38 2021 Very Low Very Low Medium Medium 68.0 0.795 74.9 18.7 9.0 13484.0
179 Malta 0.69 2.31 29.96 15.04 54.76 20.09 2021 Very Low Very Low Very Low Very Low 23.0 0.918 83.8 16.8 12.2 38884.0

70 rows × 18 columns

In [38]:
fig = px.scatter(wri_hdi_2021_cleaned, x='Human Development Index (HDI) - 2021', y='Vulnerability', 
                 title='HDI vs Vulnerability with Regression Line',
                 labels={'Human Development Index (HDI) - 2021': 'HDI', 'Vulnerability': 'Vulnerability'},
                 color_discrete_sequence=['blue'])

x = wri_hdi_2021_cleaned['Human Development Index (HDI) - 2021']
y = wri_hdi_2021_cleaned['Vulnerability']
slope, intercept = np.polyfit(x, y, 1)
regression_line = slope * x + intercept

fig.add_trace(go.Scatter(x=x, y=regression_line, mode='lines', name='Regression Line', line=dict(color='red')))

fig.update_layout(
    xaxis_title='HDI',
    yaxis_title='Vulnerability',
    showlegend=True
)

fig.show()

There is a strong negative correlation between HDI and WRI. Countries with higher levels of human development, such as better education, health, and living standards, generally experience lower levels of risk associated with natural disasters.

In [39]:
df_grouped = df.groupby(['Year', 'Continent']).mean().reset_index()

wri_pivoted_vulnerability = df_grouped.pivot(index='Year', columns='Continent', values='Vulnerability')

fig, ax = plt.subplots(figsize=(10, 6))

for region in wri_pivoted_vulnerability.columns:
    ax.plot(wri_pivoted_vulnerability.index, wri_pivoted_vulnerability[region], label=region)

ax.set_title('Exposure Over Time for All Continents')
ax.set_xlabel('Year')
ax.set_ylabel('Vulnerability')
ax.legend(title='Continent')
ax.grid(True)

plt.tight_layout()
plt.show()
In [40]:
df_grouped = df.groupby(['Year', 'Region']).mean().reset_index()

wri_pivoted_vulnerability = df_grouped.pivot(index='Year', columns='Region', values='Vulnerability')

fig, ax = plt.subplots(figsize=(10, 6))

for region in wri_pivoted_vulnerability.columns:
    ax.plot(wri_pivoted_vulnerability.index, wri_pivoted_vulnerability[region], label=region)

ax.set_title('Exposure Over Time for All Regions')
ax.set_xlabel('Year')
ax.set_ylabel('Vulnerability')
ax.legend(title='Region')
ax.grid(True)

plt.tight_layout()
plt.show()